Material guide: perilla alcohol — two of its numbers look wrong, and it is also a drug in clinical trials

CAS 536-59-4, 12 suppliers, cleared by JECFA and FEMA GRAS. It is limonene hydroxylated at C7, and it smells green, cumin, spicy, cardamom, orange peel. Its boiling point field carries 119 °C @ 11 mmHg alongside 93 °C @ 760 mmHg; a reduced-pressure boiling point cannot exceed the atmospheric one, and of the 1,258 materials registering both, only two commit that error. Its substantivity reads 269 hours, while its isomer laevo-carveol, same formula C10H16O, same molecular weight 152.23, and twice the vapour pressure, reads 51 hours. Linalool in the same family reads 12 hours, geraniol 60, citronellol 56. This one runs about four times longer than its family predicts. Outside perfumery, highly purified perillyl alcohol has been through a phase I intranasal trial in recurrent glioblastoma.

About a 5 minute read.

The short version

Name: perilla alcohol (perillyl alcohol), systematically p-mentha-1,8-dien-7-ol, CAS 536-59-4
C10H16O, molecular weight 152.24, 12 suppliers
Odour type green; odour: green, cumin, spicy, cardamom, woody, floral, violet, orange peel
Medium strength (rank 2), substantivity listed at 269 hours. I think that number is too high
The boiling point field contradicts itself: the reduced-pressure value exceeds the atmospheric one by 26 °C, which is impossible
Regulatory: JECFA, FEMA GRAS, CoE, FLAVIS. Oral LD50 2,100 mg/kg
Appears once in 954 formulas, at 7.00%
A highly purified version is in clinical trials as a nasal spray for recurrent glioblastoma

It is limonene with one hydroxyl added

The systematic name p-mentha-1,8-dien-7-ol unpacks plainly: a p-menthane skeleton, double bonds at positions 1 and 8, and a hydroxyl on carbon 7. Remove that hydroxyl and you have dextro-limonene.

Limonene came second when I measured the dose distribution, at a median of 15%. One hydroxyl turns a citrus material used by volume into a niche green-spice material with 12 suppliers.

The smell moves with it. There are two entries in odor_descriptions:

At 100%: green, cumin, spicy, aromatic, woody, cardamom, floral, waxy, violet

At 1%: sweet, woody, aromatic spicy, cardamom and green cumin-like, with dried orange peel and green waxy floral nuances

Cumin and cardamom are the easiest way to recognise it. It is not a material for building citrus.

That boiling point is impossible

The bp field reads:

119.00 to 121.00 °C. @ 11.00 mm Hg | 92.00 to 93.00 °C. @ 760.00 mm Hg

Lower pressure, lower boiling point. 11 mmHg is a seventieth of atmospheric, and a boiling point there cannot sit 26 °C above the atmospheric one.

I swept the whole database: of 1,258 materials registering both an atmospheric and a reduced-pressure boiling point, only two genuinely break this rule, and this is one of them (the other is 7-methyl coumarin).

Which value is right? Work back from the vapour pressure. Its 25 °C vapour pressure is 0.006 mmHg, and the relationship I fitted across 8,048 materials is boiling point ≈ 181.5 - 36.5·log10(vapour pressure), which gives 263 °C.

The 93 °C is far off; the usable value is 119 °C @ 11 mmHg. The 93 is most likely a measurement taken at a lower pressure that was filed under the wrong one.

269 hours is longer than anything in its family

Setting the C10 monoterpene alcohols side by side:

Material	MW	Vapour pressure @25 °C	Substantivity	Suppliers
dextro-limonene (no hydroxyl)	136.23	0.198	4 h	82
perillaldehyde	150.22	0.043	8 h	23
linalool	154.25	0.016	12 h	135
alpha-terpineol	154.25	0.028	20 h	103
dihydrocarveol	154.25	0.100	20 h	8
menthol	156.26	0.032	32 h	114
nerol	154.25	0.013	44 h	85
laevo-carveol	152.23	0.0117	51 h	16
citronellol	156.26	0.020	56 h	119
geraniol	154.25	0.021	60 h	198
perilla alcohol	152.23	0.006	269 h	12

Laevo-carveol is its isomer: the same C10H16O, the same molecular weight of 152.23. Its vapour pressure is twice as high, and its substantivity is 51 hours.

Perilla alcohol does carry the lowest vapour pressure in the group, so lasting longer is reasonable. The question is by how much. At the roughly threefold-per-decade rate I measured, a twofold vapour pressure difference is about 0.3 of a decade, which would take carveol's 51 hours to around 71.

The registered value is 269, about four times what the family predicts.

The same thing happened with the last material I wrote up: 216 hours, about ten times its family's prediction. Two consecutive materials overshooting at the long end may not be coincidence, but two cases will not carry a conclusion.

In practice I plan around a material that lasts tens of hours, not 269.

It is also a drug in clinical trials

The notes field carries content you rarely see in a fragrance database:

Inhibits geranylgeranyl transferase. Perillyl alcohol is a monoterpene isolated from the essential oils of lavandin, peppermint, spearmint, cherries, celery seeds and several other plants. In animal studies it has been shown to regress pancreatic, mammary and liver tumors.

That is not overstated. Highly purified perillyl alcohol (NEO100) completed a phase I intranasal trial in adults with recurrent glioblastoma in 2021 (PMID 33604574). The nasal route is there to bypass the blood-brain barrier.

None of this bears on your formula, but it explains two things: why a niche material with 12 suppliers carries such complete toxicology and regulatory fields, and why its cosmetic_uses field lists antimicrobial agents and antioxidants alongside fragrance.

On the other side, Alonso-Gutierrez and colleagues reported metabolic engineering of E. coli in Metab Eng in 2013 (PMID 23727191), building limonene and perillyl alcohol production into microbes. The same molecule sits in a drug trial on one side and a fermentation process on the other.

Where it occurs in nature

The occurrence field lists many sources, all at low concentration:

Source	Content
Amomum testaceum fruit oil (Malaysia)	1.70%
Angelica root oil	0.10%
Caraway seed oil	0.10%
Mandarin oil (Uruguay)	0.02–0.03%
Lavandin, bay laurel leaf, ginger, orange peel	not quantified

The highest is a fruit oil from the ginger family at 1.70%, which is the same fact as its smelling of cardamom.

Dose: one formula in the corpus

It appears once in the 954 formulas, at 7.00%.

One formula yields no median. I list it to make a different point: this material is essentially absent from the demonstration corpus. Strength rank 2, a substantivity of 269 hours on the record, every regulatory clearance in place, and still only 12 suppliers and a single formula.

If you are looking for a material nobody uses, this is one. If you are looking for a safe bet, it is not.

What this doesn't establish

I measured neither the boiling point nor the substantivity. Both judgements are internal consistency: the boiling point breaks the monotonic relationship between pressure and boiling point, and the substantivity departs from its family's vapour pressure trend. That is enough to say those two numbers are untrustworthy, not enough to supply correct ones.
The 263 °C is a regression estimate with a residual standard deviation of 34 °C. Treat it as an order of magnitude, not a measurement.
The family comparison holds only 11 materials. I picked those C10 monoterpene derivatives by hand; it is not a random sample.
Two consecutive long-substantivity overshoots are not evidence. Claiming the substantivity field systematically over-reports at the long end would need a distribution comparison across the whole field, which I have not run.
The clinical trial has nothing to do with perfumery. I included it because it explains why this record is so complete, not to suggest the material has any therapeutic effect. A phase I trial measures safety and dose, not efficacy.
The assay field reads 96.00 to 100.00, which is not a default template value, but neither does it tell you what your lot actually is. Read the supplier's certificate of analysis.